[Bugfix] Key CUDA graph dedup signatures on kernel function identity (#37657)
This commit is contained in:
@@ -107,6 +107,7 @@ def kernel_node_payload(node):
|
|||||||
params = checkCudaErrors(cuda_drv.cuGraphKernelNodeGetParams(node))
|
params = checkCudaErrors(cuda_drv.cuGraphKernelNodeGetParams(node))
|
||||||
return (
|
return (
|
||||||
kernel_name(params),
|
kernel_name(params),
|
||||||
|
(int(params.kern), int(params.func)),
|
||||||
(int(params.gridDimX), int(params.gridDimY), int(params.gridDimZ)),
|
(int(params.gridDimX), int(params.gridDimY), int(params.gridDimZ)),
|
||||||
(int(params.blockDimX), int(params.blockDimY), int(params.blockDimZ)),
|
(int(params.blockDimX), int(params.blockDimY), int(params.blockDimZ)),
|
||||||
int(params.sharedMemBytes),
|
int(params.sharedMemBytes),
|
||||||
|
|||||||
Reference in New Issue
Block a user